Matrox Parhelia Launch Review - PAGE 7

How much did matrox pay you guys to right this review? Every other site has the card running 40 fps less then you.

How on earth did you guys get such a faster card??!?!

Seriously, why are all the other review sites posting much lower benchmarks than you are for the Parhelia? Belive me, if I can buy the Parhelia off the shelf and see the exact performance you're getting here or better, than heck, that's good enough for me! Especially if they have good Linux drivers. Do you know what core clock and memory speed your card has for sure?

I haven't checked the other sites for benchmarks yet, but I saw Neumann running the benchmarks at our office - he ran the same benchmarks he normally does and he walked away from the system while it was running through everything.

Is there really a 40fps difference??

Can you clarify what you mean by our numbers are higher then anyone else's numbers?

I'm doing a side by side comparison between our review and HotHardware's review.

Look at their MadOnion Scores here:
http://www.hothardware.com/hh_files/S&V/parhelia128mbtest(5).shtml

It mirrors exactly our results, which show the Parhelia trailing the 8500, which trails the Ti4600 by a lesser margin.

Now when I compare the Q3 timedemo scores here:
http://www.hothardware.com/hh_files/S&V/parhelia128mbtest(4).shtml

I see the same trend with our results.

Our results are here btw:
http://www.neoseeker.com/Articles/Hardware/Reviews/parhelia/6.html

The thing I do notice the most is that in the Q3A time demos, we're showing less difference in speed between the the Parhelia and the GF4 - but I think this can be attributed to raw CPU power.

Our disclosed setup:

AMD XP1800 1.53GHz
ECS K7S5A SiS735 w/ Ultratimings
512MB Micron PC2100 DDR
Maxtor 60GB 7200RPM
ATI 8500 7.72-020524M-004273C Catalyst 2.1
Matrox Parhelia-512 1.0023 Drivers
Matrox Parhelia-512 1.0025 Drivers
Windows XP Professional SP1

I'm guessing that the XP1800 is just not powerful enough to show the larger difference in performance - so a more powerful chip would have been more clarifying here - that's sort of a logistics thing since Neumann didn't have a 2000+ or faster chip on him at the time.

That may mean that the review is imperfect because we don't have the best equipment, but it hardly means we're being paid off. The trends reflect the same things that other sites have found.

And why the HECK would a site lie about benchmarks when 10 other sites will have a review on the same day or withing the next week or so?

I would say the the review is informative, and I have no reason to doubt its accuracy. Most reviews that I have seen have been running stupidly fast computers like a Northwood 2.53 P4 which I very much doubt that a huge proportion of home users have. I would suspect that there are a lot more of us out there that have Athlon XP's around the 1800+ (1.53Ghz) mark. This could well show a slightly different picture due to raw processor power, but you have to remember that the Parhelia is clocked an awful lot slower than the GeForce4 Ti. If Matrox put out a card clocked at a similar speed we might see a slightly different story.

I would say that this is a good review that gives most people an idea of the sort of power they will get on thier setup at home, and from what I can see at high resolutions, with swanky GFX features like FSAA the Parhelia is definetly worth it, lets just hope the clock speed of the board is boosted in the near future to make it try and keep up with those more powerful boards at lower resolutions/without FSAA.

-----------------------------
Fyrespray

http://www.ukgamer.net
http://www.ukgamer.net/columns.php3?author=9

quote Redemption

Can you clarify what you mean by our numbers are higher then anyone else's numbers?

Sure, I'll give some examples.

quote

Now when I compare the Q3 timedemo scores here:
http://www.hothardware.com/hh_files/S&V/parhelia128mbtest(4).shtml

I see the same trend with our results.

Our results are here btw:
http://www.neoseeker.com/Articles/Hardware/Reviews/parhelia/6.html

The thing I do notice the most is that in the Q3A time demos, we're showing less difference in speed between the the Parhelia and the GF4 - but I think this can be attributed to raw CPU power.

I'm guessing that the XP1800 is just not powerful enough to show the larger difference in performance - so a more powerful chip would have been more clarifying here - that's sort of a logistics thing since Neumann didn't have a 2000+ or faster chip on him at the time.

Well, for example, the Q3 timedemos here:

http://www17.tomshardware.com/graphic/02q2/020625/parhelia-13.html

At this point, that's the only thing I can guess, would it be possible to use Quake 3 1.17 to do the benchmarking like the majority of other sites, to see if that is what the performance difference is? The CPU must be the other difference, Tom's used a Pentium 4 2.2GHZ for their review. It's just odd that in your review on an Athlon XP 1800+ a Radeon 8500 gets 129fps at 1024x768, and the Parhelia roughly the same (1fps difference), while at THG there's a 50fps difference between the two cards! This is what's so startling.

quote

And why the HECK would a site lie about benchmarks when 10 other sites will have a review on the same day or withing the next week or so?

I never said that you lied. In fact you'll note that I stated that if the performance difference was really that little, then that would certainly be good enough for me to use that as a justification to buy the card. I only asked why on earth other sites seemed to have much lower scores than you. In fact, I happen to own a system at home that's just about identical to the test one used in your article. (Athlon XP 1800+ system)

So again, please don't take this as a slight, it's just *many* users on the matroxuser fan sites and the like all wondered why the picture seemed much more rosy with your review. It seems as large differences in test setup may be to blame.

quote EvilTypeGuy

Well, for example, the Q3 timedemos here:

http://www17.tomshardware.com/graphic/02q2/020625/parhelia-13.html

At this point, that's the only thing I can guess, would it be possible to use Quake 3 1.17 to do the benchmarking like the majority of other sites, to see if that is what the performance difference is? The CPU must be the other difference, Tom's used a Pentium 4 2.2GHZ for their review. It's just odd that in your review on an Athlon XP 1800+ a Radeon 8500 gets 129fps at 1024x768, and the Parhelia roughly the same (1fps difference), while at THG there's a 50fps difference between the two cards! This is what's so startling.

Ok I can see where some confusion comes from. We were using a RADEON 8500 with 64MB of RAM. This is an extremely important point because it's not exactly a fair comparison - BUT - note that the GF4 is still not performing that great on the 1800XP (ie it's lead over the Parhelia is not that great), so I'm still thinking this is a CPU dependent situation. As for why we used a 64MB RADEON - that's all we had on hand at the time - and Matrox never let us know what the memory configuration was going to be for our test unit - we had 3 business days between the time we got the Parehlia and when the NDA expired. This sorta left us little time to get ATI to ship us a card for a proper comparison. I think we're going to ask ATI to get us a 128MB card to do a rematch - if time permits.

quote EvilTypeGuy

I never said that you lied.

Actually, that reference about why we would lie was referring to the comment made by Avail about being paid by Matrox.

Hi all
I see this thread is building
I was my decision to use Quake 3 Arena v1.31 retail being the newest release by idSoftware. A lot of other websites use either the demo version or can't convert the demo files from protocol 66 to 67. I'll check in with idSoftware about the differences with v1.17 versus v1.31 but i have a feeling they will encourage me to use the new release...??

I really don't see why people think this review is biased in anyway. Is it the fact the other sites were using KT333 chipsets versus my old SiS735 chipset? The fact that they were using a lot of games as benchmarks for a card obviously not geared towards gaming??

Really, the Parhelia is a innovative new graphics card and seriously the greatest features it has are 10bit colouring and multi-display surround gaming. I would really like to see more products based on this technology in the future from either ATI or nVidia.

on a further note... if only i was being paid to do these reviews...sigh;(

i wanna see the tests done on a old school celeron (it must be this chip cause it doesn't have sse sse2, and 3d-now instructions, with 384mb sdram, no tweaks what so ever, and on win me and xp
with quake3 running at 1024 by 768 32bit color using only the original drivers that comes with the cards and the max setting for antialising must be on as well as the max filtering

i wonder which card will come out top then

I have the card and running the MadOnion 3dMark2001 it performs only 30% better than a GeForce 2 MX with 32Mb of memory. The one thing though that looks like an issue is that the report says I have only 16Mb total AGP memory. I do not know whet this means or if it is an issue. Can anyone comment?